Promise SanLink2

[Barry Sharp]

I suspect the extra 20% CPU use is caused by the Promise SANLink2 driver software as it’s shown as kernel CPU use.

[Barry Sharp]

I made a further check into the use of SANLink2 for my purposes/goals and used the SANLink Utility to spiff up the quality of the data movements. With this I came close to what the Thunderbolt Bridge delivered. No matter, the Thunderbolt Bridge still trumped the use of SANLink2 Adapters for my case.

This has been a good educational task and now I can put it behind me and return the SANLink2 Adapters for a full refund. I must say the SANLink2 Adapters are very well made, are quiet, and run a bit warm to touch. The Adapters having two 10G ports is a bonus for sure and could be used for bonding for link aggregation, improved bandwidth and fail-over protection for one link going down/failing. They are costly but much less (1/2 the cost) than the earlier PCIe 10G cards of the past for older MacPros. If and when our office require the need for a true File Server, and we still have the MacPros with Thunderbolt ports, the likes of the SANLink2 Adapters will be high on our list.

The bottom line is that we can enjoy our $60 Thunderbolt Bridge solution knowing that 10G brings little to the table for helping us at close to 25x our current cost.

[Bob Zelin]

Hi Barry –
if you don’t mind me asking (perhaps this was mentioned earlier in this LONG thread) – I am under the impression that you are using Thunderbolt bridge networking via AFP to connect to Mac Servers in the office where editing is done.
May I ask you – exactly how many client computers are there, and exactly what drive array are you using in this configuration that you state works perfectly fine for video editing ?

Thank you very much (and sorry for driving you crazy with this).

Bob Zelin

Bob Zelin
Rescue 1, Inc.
bobzelin@icloud.com

[Barry Sharp]

Bob:

Yes I did detail our office setup (somewhat) in an earlier posting, but maybe I did not clearly describe it or it was incomplete.

Today we have three Macs. We do also rent extra Macs at times when Project workload demands it.

1)
Pseudo file server; MacPro6,1, (2.7GHz 12core, 1TB SSD, 64GB RAM, Dual D700s)

This is the pseudo file server in our central office. It has an 8 disk SoftRAID managed RAID-5 made up with 6x 2TB and 2x 3TB Toshiba 7200 rpm disks accessed over Thunderbolt. We use SoftRAID version 5.5.5.

The RAID-5 is partitioned into 4x 3.5TB (known as BIG-1, BIG-2, BIG-3 and BIG-4) and 2x 1TB JBODs for OS backups (Primary & Test OSes).

The BIG-1/2/3/4 can deliver sustained 650 MBytes/sec for reads and some sustained 350 MBytes/sec for writes. These units are use for loading up Project data to be used by all office Macs.

The BIG-1/2/3/4 units are backed up daily in early morning hours to a MacGurus 5-bay Burly eSATA Port Multiplier that has 5x 4TB disks. The Burly unit is daisy chained off the 8 disk RAID-5 unit using a TSATAII-PRO-E34 – Sonnet Tempo 2 Port Express34 Pro SATA Host Card in a Sonnet ECHOPRO-E34 – Sonnet Echo ExpressCard Pro.

This MP6,1 is also used for all Project workloads.

2)
Client MP6,1 (3.5 GHz 6core, 1TB SSD, 64GB RAM, Dual D700s). Has a local 4T USB3 unit for local backups. This MP has Thunderbolt Bridge connection the the pseudo file server in our central office.

3)
Client MP6,1 (3.5 GHz 6core, 1TB SSD, 64GB RAM, Dual D700s). Has a local 4T USB3 unit for local backups. This MP has Thunderbolt Bridge connection the the pseudo file server in our central office.

Today all the Mac systems run latest version of Apple’s El Capitan. We are testing macOS Sierra at this time when we have some free time over weekends.

All three Macs use bonded 1GbE ethernet to a managed Cisco SG200-18 switch.

Sometimes we bring in a high-end 27″ iMac to help with workload. We may use Thunderbolt Bridge or bonded 1GbE ethernet to connect it to the pseudo file server in our central office via the Cisco SG200-18 switch.

Each client and the pseudo file server will read their respective piece of the current Project and crunch away independently and return their results to the pseudo file server when done. There is never any confusion with file collisions happening on the pseudo file server. Original Project data is never ever overwritten. The pseudo file server Mac is responsible for pulling all the Client results together to produce final results.

The one weakness in our setup is if the pseudo file server hangs or crashes during heavy Project workloads. This has been a problem for us as now and then as Mac OS can act up at times as can the Adobe/Premier/FCP software etc.

Our MP6,1 systems run 24/7 and have been for nearly 3 years now without any hardware issues at all. The Burly did fail one time and I had to replace its power supply and install some stealthy fans (very quiet). The 8 disk RAID-5 unit lost one of its disks just a month ago but being a RAID-5 no data was lost.

If we were to grow much beyond having 4 Macs we would definitely need a bone fide File Server than can deliver at least ~400 MBytes/sec to each and every Client machine at same time. We would at this point abandon the use of Thunderbolt Bridge because it would become too unwieldy and quite likely there would be insufficient Thunderbolt ports for connections. Thus if say we had 5 Macs in the office a file server would need to push out at least 2000 MBytes/sec at any given time over multiple 10G connections. This would be a nice and easily manage network but obvious would require some serious brass/money. ?

I hope the above helps.

[Chris Crichlow]

Hi All,

Hope all is well!

This forum has been a pretty solid well of knowledge for me when trying to work out a little conundrum of my own. Wondering if you guys can help!

I’m currently looking after a few event spaces and 2 edit suites. I’m having crazy issues with FCPX (v10.4) on my two edit machines, linked back to an EVO RAID5 NAS. My edit suites are running Mac Pros (10.13.4, 2.7Ghz 12 Core Intel Xeon E5, 64GB 1866Mhz DDR3), connected to a SanLink2 (all connected interfaces are running on High Throughput) which then leads to an EVO RAID5 NAS over fibre (10GbaseSR, 9000MTU).

Bear with me… This part becomes a little convoluted. The edit suites are fed video files from two other MacPros of the same spec which are ingesting live 1080p/50 3G-SDI feeds, converting them over an AJA I/O 4K interface to thunderbolt using Softron MovieRecorder. The feeds are then pushed down as ProRes LT to the EVO servers over SanLink2 units.

The EVO is segmented into various partitions – one of which houses the local files and the other partitions are for the edit suites (projects storage, cache, etc).

Since our upgrade to High Sierra, and no other infrastructure changes, FCPX has been lagging and beachballing beyond belief! I get continual hangs and regular beachballing when editing and I’m totally unable to get to the bottom of it.

I’ve attached my AJA System Test for one partition, the rest only vary by 20 -/+ MB. At this point, I’m at a bit of a dead end though. Any ideas anyone?

Hope you can help!

Thanks

Chris

[Bob Zelin]

Hi Chris –
I have no idea of why you added to a post that started in 2014, and ended in 2016. You should have started a new post for this.

1) . You own a SNS EVO. Why don’t you call SNS for support ?
2) you upgraded to High Sierra. You need to turn off SMB signing on the Macs, and you need to reinstall the
SanLink2 drivers, which you will not be able to do unless you turn off Apple System Integrity Protection, which will block you from installing the Promise drivers (the simple “click Allow” in Security and Privacy no longer works.

Bob Zelin

Bob Zelin
Rescue 1, Inc.
bobzelin@icloud.com

[Chris Crichlow]

Hi Bob,

Thanks for the response. I was just adding to it as it was relative to the discussion that was going on, which various others had contributed to. No problem for future postings.

Prior to calling SNS, I wanted to see if there were things that others had tried that I hadn’t. My SMB signing is already off and I have the drivers installed, this was one of the first things I did in setup of the system to enable the connection. What I’m trying to work out is where the throttling is happening and why this is affecting FCPX so heavily. My colleagues at Apple didn’t have an answer, so I thought I’d reach out to various people in the pro community who use these interfaces in their workflows.

I’ll direct my query to SNS.

Thanks.

C

[Barry Sharp]

I don’t see the attached AJA System Test for one partition! Maybe I’m not looking in the right place for it….

[Barry Sharp]

Do you see the same issue when JUST one FCPX is running and pulling data from the EVO ?

[Barry Sharp]

Did you first remove the previous/old SanLink2 driver code before installing the new code ?